Qiu Rui, Wang Jiaxin, Ren Qinqi, Huang Weihong, Zhu Jiahao, Liu Dexing, Gao Xinyu, Wang Wanting, Liu Qi, Zhang Min
School of Electronic and Computer Engineering, Peking University, Shenzhen 518055, China.
ACS Nano. 2023 Jul 11;17(13):12652-12662. doi: 10.1021/acsnano.3c03212. Epub 2023 Jun 29.
The coexistence and interaction of excitatory and inhibitory neurotransmitters at biological synapses enable bilingual communication, serving as a physiological foundation for organism adaptation, internal stability, and regulation of behavior and emotions in mammals. Neuromorphic electronics are expected to emulate the bilingual functions of the biological nervous system for artificial neurorobotics and neurorehabilitation. Here, we have proposed a bilingual bidirectional artificial neuristor array, which utilizes ion migration and electrostatic coupling properties between intrinsically stretchable and self-healing poly(urea-urethane) elastomer and carbon nanotube electrodes, realized by van der Waals integration. The neuristor exhibits depression or potentiation behaviors in response to the same stimulus in different operational phases and achieves a four-quadrant information-processing capability. These properties make it possible to simulate complex neuromorphic processes, which involve bilingual bidirectional responses, such as withdrawal or addiction responses, and array-based automated refresh. Furthermore, the neuristor array is a self-healing neuromorphic electronic device that can function effectively even under 50% mechanical strain and can recover operation voluntarily within 2 h after experiencing mechanical injury. Additionally, the bilingual bidirectional stretchable self-healing neuristor can emulate coordinated neural signal transmission from the motor cortex to muscles and integrate proprioception through strain modulation, similar to the biological muscle spindle. The properties, structure, operation mechanisms, and neurologically integrated functions of the proposed neuristor signify an advancement in neuromorphic electronics for next-generation neurorehabilitation and neurorobotics.
兴奋性和抑制性神经递质在生物突触处的共存与相互作用实现了双向通信，为哺乳动物的机体适应、内部稳定以及行为和情绪调节提供了生理基础。神经形态电子学有望模拟生物神经系统的双向功能，用于人工神经机器人和神经康复。在此，我们提出了一种双向人工神经晶体管阵列，它利用了本征可拉伸且自愈合的聚（脲 - 聚氨酯）弹性体与碳纳米管电极之间的离子迁移和静电耦合特性，通过范德华集成实现。该神经晶体管在不同工作阶段对相同刺激表现出抑制或增强行为，并实现了四象限信息处理能力。这些特性使得模拟复杂的神经形态过程成为可能，这些过程涉及双向响应，如退缩或成瘾反应以及基于阵列的自动刷新。此外，该神经晶体管阵列是一种自愈合神经形态电子器件，即使在50%的机械应变下也能有效工作，并且在经历机械损伤后能在2小时内自动恢复运行。此外，这种双向可拉伸自愈合神经晶体管能够模拟从运动皮层到肌肉的协调神经信号传输，并通过应变调制整合本体感觉，类似于生物肌梭。所提出的神经晶体管的特性、结构、运行机制和神经整合功能标志着神经形态电子学在下一代神经康复和神经机器人领域取得了进展。
